Snag the bug,
ship the fix.
Your users catch a bug in one click. Snag captures the replay, console, and network, files it into Jira + GitHub, and lets a coding agent open the fix PR.
No signup needed — walk through a real captured bug in about a minute. See it work
From “it’s broken” to a reviewable PR.
Existing tools stop at filing the bug. Snag closes the loop to a pull request.
Capture in one click
Reporter hits the extension on an authorized domain. Replay, console, network, and a repro summary — masked in-browser.
Route with full context
A complete ticket lands in Jira with a replay link, linked to the right GitHub repo. Triaged and deduped automatically.
Open the fix PR
Your agent — or Snag's — reads the context, analyzes the repo, and opens a PR for a human to review. Never auto-merged.
Every signal, in one screen.
Session replay synced to the console, network, and repro timeline — plus the agent’s proposed diff. The mono surfaces developers live in, done right.
Plug in your agent, or use ours.
The same capture pipeline and context payload feed both — only the “who touches the code” boundary moves.
Your agent, your credentials
Snag exposes ticket context over MCP. Cursor, Claude Code, or Copilot reads exactly what it needs and opens the PR. No source ever leaves your org.
Listed in the official Model Context Protocol registry as ac.snag/snag.
Turnkey, sandboxed
Snag’s agent analyzes your repo in an isolated, ephemeral sandbox and opens a least-privilege PR. Source is destroyed after every run; models never train on your code.
Masked by default
Passwords and inputs masked, headers stripped, secrets redacted — in the browser, before upload.
Scoped & verified
The extension only records on domains you verify, authenticated by a hashed workspace key.
Human in the loop
The agent opens a PR for review. Snag never auto-merges and never trains on your code.
More on how this works on the security page.
Flat pricing. Unlimited members. Pay for fixes.
No per-seat math — invite your whole team and let anyone report a bug. Fix plans include a monthly pool of fix-runs; go over and you only pay for the fixes you run. Monthly only — cancel anytime, prorated automatically.
- Free — 15 captures / month Try capture on your own site. Start free
- Capture — $29/mo. Unlimited capture + MCP Teams bringing their own agent — Cursor, Claude Code, Copilot. Start Capture
- Starter — $49/mo. Includes 15 fix-runs/mo. $5 per fix-run after A small team trying the built-in agent. Start Starter
- Team Recommended — $99/mo. Includes 40 fix-runs/mo. $5 per fix-run after A product team fixing bugs every week. Start Team
- Business — $299/mo. Includes 150 fix-runs/mo. $5 per fix-run after Several teams that need roles and priority support. Start Business
- Enterprise — Custom
Capture is $29/mo for your whole workspace — about what Jam charges for two people.
What’s a fix-run?
Capture and routing are unlimited on every paid plan — and free for life for the founding 50 workspaces (the Free plan includes 15 captures a month). You only spend a fix-run when Snag’s AI actually diagnoses a bug and opens a pull request. A run that doesn’t end in a PR costs half — you pay for work products, not attempts. (Trial and referral runs come back in full.)
Unlimited members and reporters on every plan; capture is unlimited from the Capture plan up (Free includes 15/month; founding workspaces are unlimited for life). 1 fix-run = one built-in-agent diagnose-and-PR attempt; overage is billed per fix-run, and a run that doesn't end in a PR costs half (trial and referral runs come back in full). Cancel anytime — you're credited the prorated amount for unused time. Payments are processed by Stripe — Snag never sees your card details. Your first 5 fixes are on us, and no credit card is required to start.
Questions teams ask before they install
Is my source code safe in the sandbox?
The built-in agent works on a disposable copy of your repo in an isolated, ephemeral sandbox, and that copy is destroyed after every run. Snag keeps the bug evidence and the PR history, never your source, and nothing is used to train models. The details are on the security page.
What happens when a fix-run fails?
The ticket comes back to your team with everything the agent found: the evidence, where it looked and why it stopped. A run that doesn't end in a PR costs half a fix-run, and a trial run costs nothing.
Does it work with Cursor or Claude Code?
Yes. From the Capture plan up, Snag exposes the captured evidence over MCP, so Cursor, Claude Code or Copilot can read it and open the PR themselves. The tool reference is at snag.ac/mcp.
Can I really have unlimited reporters on a flat plan?
Yes. Every plan has unlimited members and reporters: you pay per workspace, not per seat. Capture is unlimited from the Capture plan up, and the Free plan includes 15 captures a month.
Which languages can the fix-run work in?
Node.js, TypeScript, Python, Ruby, PHP, Java and Kotlin, .NET, Go and Rust. Snag detects your stack from the repo and builds the sandbox to match, so your tests run before the PR opens. On anything else the agent still reads and searches your code and proposes the fix; it just can't run your tests in the sandbox yet.
Who is behind Snag?
Snag is built by Drew Stockler and operated by Digital Ventures Conglomerate Limited. Why it exists, in the founder's own words.
Stop chasing repros. Start shipping fixes.
Install in minutes. First 50 workspaces: lifetime free capture + MCP.